A journey through words and numbers to a qualitative understanding of need

Paper presented at the Association for Qualitative Research International Conference, Melbourne, July 6-10, 1999.

Lynn Kemp

Remedial and Research Solutions
4 Cessna Place
RABY NSW 2566
Ph: 02 9820 8613

Introduction

An understanding of the community service needs of persons with spinal injuries has been developed in the first comprehensive investigation of persons with spinal injuries in New South Wales. The study aimed to understand need and provide guidelines to policy makers as to how to address the community service needs of persons with spinal injuries. Different concepts of need were explored:

• Normative need – what the expert professional perceived to be needed based upon population norms (Bradshaw, 1972);
• Felt need – perceived by the individual (Bradshaw, 1972);
• Expressed need – felt need turned into action in the form of a request for a commodity (Bradshaw, 1972);
• Prescriptive need – the resources required to make good deficits in the provision of a commodity (Thayer, 1973);
• Comparative need – deduced by an outside observer in circumstances where individuals not in receipt of a particular commodity have similar characteristics to others who do receive it (Bradshaw, 1972);
• Intrinsic need – where a needed commodity is both a means and an end, that is, having some desired thing is an intrinsic part of in some desired state (Miller, 1976), and
• Need as a means to an end – need 'defined as those things which are necessary for the formulation and execution of plans and purposes which are self-determined' (Stainton, 1994 p.115).

Exploring different concepts of need required multiple methodologies including:

• a review of government policies and legislation,
• statistical and segmentation analyses (SPSS CHAID, Magidson, 1993) of demographic, injury related and service usage data obtained from a census-type survey, and
• conceptual and grounded theory analysis of interviews with persons with spinal injuries and their carers.

To address the assessment of the needs of poorly identified groups and bridge the gap between macro, or population based understandings of need, and micro assessment of individuals' felt needs, a number of health researchers in the United Kingdom have advocated the use of multiple methodologies combining qualitative and quantitative methods. Suggested methods include the use of scales to determine the importance of needs, in-depth interviews, and the identification of individual needs through consultation (Endacott, 1997). Such triangulation will, it is suggested, give a well-rounded picture of a population's needs for care, an indication for provision, and present results highlighting mis-matches between need and…care delivery (Billings & Cowley, 1995 p.723).

Whilst commencing my research with the intention of undertaking triangulation to achieve a well-rounded picture, I must admit I was perplexed by the mixed metaphors – a feeling which increased as the research process unfolded. In order to understand my consternation, today's journey begins with the concept of triangulation.

Triangulation and crystallisation

Triangulation is most often referred to as the use of more than one method in a study with the goals of achieving confirmation and/or completeness, and is most often currently used as a synonym for a particular form of triangulation called by Denzin (1978) 'between methods', 'methodological triangulation'. Triangulation, usually described, is an apt term because there is a distinct sense of linearity in the process. The 'blending' of qualitative and quantitative methods can be sequential – qualitative study preceding and informing the development of quantitative instruments, or qualitative study forming a postscript to a quantitative study – where each stage of the sequence is a 'new' study – confirming or rounding out the previous findings. Alternatively, triangulation can occur in a concurrent fashion – a kind of jigsaw where you believe that the use of multiple methods will show you various parts of the picture and you can put it together as a whole.

This is, admittedly, a very simplistic view of triangulation; combinations of the above can result in a complexity of form. It is not my intention, however, in this paper to give an extensive lesson on triangulation. It is my intention to point out to you the linearity of process within triangulation which, whilst providing a 'complete' picture, prevents any alteration of the component parts. Once you have decided on the shape of the triangulation, the methods become linear, rigid and unchangeable.

More recently, the metaphor of triangulation has been expressed as crystallisation (Richardson, 1994 cited in Proctor, 1998). The proponents of this metaphor highlight the increased complexity of the crystal and its ability to change appearance depending upon the angle of observation. Whilst I am attracted to this metaphor, a study of crystals reveals that they too have a rigidity of shape and form, made up of linear planes.

The linearity of triangulation and crystallisation is their limitation. There is a somewhat inherent positivism in the use of these methods, where one can ask a question, and by proceeding in the 'straight line' pursuit of a number of methods, the entity will become known in some way (Begley, 1996). I have found, however, that such methods are guilty of not making the most of the data, and, further, that you cannot answer some questions in this linear fashion. An answer to complex questions (such as what does someone need) can only by found using a more complex ‘blending’ or integration of methods in which quantitative and qualitative data and analyses are dependent on, inform, and become each other.

Undertaking such a process is necessarily qualitative and highly interpretive, but also extremely rigorous (I will discuss the issue of rigour later). This is true whether dealing with words or numbers. Researchers who analyse words are generally comfortable with the importance of interpretation. Researchers who use numbers are generally not. Numbers are given a special place in our society – the language of mathematics is considered to be pure and not open to interpretation, however, Kritzer, when writing about quantitative data, argues that;

To be interesting and useful for social inquiry, data must be interpreted. Data seldom speak unless asked. Many analysts often will say to themselves or others, "what are these data telling me" or "these data say to me…." (Kritzer, 1996).

Such language is very familiar to those of us who have undertaken courses in qualitative data analysis. I thus have no paradigmatic problems with 'blending' qualitative and quantitative data, as all analysis is inherently qualitative (with the exception of "brute data" such as the numerical tabulation of the number of people in a room). The study I am presenting today is an example of 'blending' of data, beyond sequential or concurrent collection and analysis, to mutation and permutation. Indeed, this paper might more rightly be entitled "the genetics of a mutating multi-method study".

Mutating methods

The study commenced using mixed methods in both a somewhat sequential and complimentary method, collecting quantitative data from over 700 people with spinal injuries across New South Wales, and collecting qualitative (written comments and interview) data from 290 of the 700. Government and service provider legislation and policies were collected and community service providers were interviewed.

The methods were applied sequentially in that the act of seeking the quantitative data temporally preceded the qualitative data collection, and the demographic data collected in the questionnaires were used to 'stratify' a sample for follow-up interviews. The methods were applied in a complimentary way in that I always intended to collect both quantitative and qualitative data and sought a 'well-rounded picture'.

From the multiple sources there were data about who the spinal injured population are, where they live, how they live, what they do in their day, how they came to have a spinal injury, what services they used, what services they wanted, how good the service are, what their rights are, what they are entitled to, how decisions about services are made, how the government provides services, how they go about getting services, how they go about using services, and much more, all available for analysis.

In keeping with the linear concept of triangulation, each individual method was pursued to their conclusion, and, as Endacott suggested, a well-rounded picture of the 'mis-matches between need and…care delivery' certainly emerged. Unfortunately, the picture I was presented with looked a bit like modern art – the pieces did not properly match. For example, I certainly was painted a picture from the quantitative data of a desperate shortage of community services, particularly for people with more significant disabilities. But I also had a picture, based upon the qualitative data, of people feeling ambivalent about the services they had most complained about not having. I had rigorously followed each method through to its conclusion, but the results were completely dissonant, and I still did not know what the spinal injured population of New South Wales needed.

The positivist, linear approach of triangulation, however, provided no answer to my research question, yet I could not stop and say my work was done. Government agencies, service providers, consumer advocacy organisations and others have been conducting multi-method studies of community service needs for nearly two decades, which have failed to correctly identify the needs of disabled persons, and resulted in little perceived improvement in service provision. Clearly, a more exploratory journey was needed.

The solution I found was to unravel the analysis, transfer the qualitative and quantitative data, and rebuild the analysis, in much the same way that genetic material reconstructs itself. The double helix of DNA comprise of a sense and an anti-sense strand, which twist around each other, and engaged in processes of mutation and protein transfer, in order to build a functionally cohesive organism. The synthesis or reconstruction of DNA molecules can be described this way:

The synthesis of a strand of messenger, transfer, or ribosomal RNA from a DNA template is called transcription. The helical DNA molecule unwinds, leaving the sense strand (the sequence from which the RNA is assembled) accessible. The enzyme that controls the reaction recognizes a "start" region, called the promoter, in the DNA sequence and builds from there. (Microsoft Encarta97).

What this meant, in this study, was that I unwound the quantitative analysis of the need for services and, using the qualitative analysis as the promoter, I rebuilt the analysis. Just as in DNA, I had two somewhat independent strands – the quantitative and qualitative analysis. The first step in rebuilding to produce a functioning organism (in this case a coherent understanding of need) was to find the sense strand.

Based upon the literature and the qualitative insights gained from the study, I returned to the analysis that made sense – there is a quantitatively demonstrable shortage of services for persons with spinal injuries. I then took the 'anti-sense', counter-intuitive qualitative data, which had been coded using QSR NUD*IST4, and imported the codes reflecting feelings about services into the project's SPSS database. This permutated data was used, together with the qualitative insights to reconstruct the quantitative analysis.

The data were used to create a variable that reflected both the current use of services, and the desire for services. Subsequent quantitative analyses revealed that the qualitative ambivalence to services was the respondents' response to the 'fact' that community services were distributed to the spinal injured population of New South Wales in an apparently arbitrary manner, rather than based upon their need for the service when defined solely quantitatively or qualitatively.

Reconstruction of the qualitative analysis then proceeded by adding the permutated data to the NUD*IST4 database for inclusion in the qualitative analysis. When reanalysed qualitatively, the quantitative arbitrariness of service provision, was, in fact, not so arbitrary. Service were allocated on the proviso that persons with spinal injuries adopt life plans which met the expectations of service providers, demonstrated by being 'just right' (for example, not being too independent, nor too dependent by demonstrating suitable levels of gratitude and humility).

Finally, it was resolved that persons with spinal injuries' need for community services could only be understood on the basis of the contribution services made to the realisation of their plans of life. Persons with spinal injuries' plans of life are to be ordinary. Ordinariness – a social normality – has little to do with level of impairment, or the other objectifiable characteristics which formed the basis for the original quantitative analysis and which are commonly used by providers to determine levels of service provision. Ordinariness is about interdependence, relationships, participation and not being special or different and, in that, was different to the rights-based, 'squeaky wheel' qualitative assumptions of service providers.

Mutating data in such a way, I may be accused of, at worst, fiddling data to produce a result, or, at the very least, conducting poorly planned research. Was I lax in my first conduct of the quantitative and qualitative analysis? Should not I have known to collect quantitative data about peoples' feelings towards services and combine the particular quantitative variables together in the particular way that produced the result? Did the discordant analysis result from a lack of rigour?

Rigour

Certainly, there are many who would argue that reconstructing and reinterpreting quantitative data in the way I have done lacks rigour. Kritzer (1996), however, argues in support of the important role interpretation plays in quantitative analysis:

In qualitative social science, the analyst must construct the text for interpretation. In quantitative social science, the analyst constructs both a first order text (in assembling the data) and a second order text (in the form of statistical results). With each additional step in the process, the role of interpretation increases…Thus, rather than being more divorced from the human process of interpretation, quantitative social science probably involves more levels of interpretation than does qualitative social science.

Qualitative methodologists, particularly grounded theorists, are much more comfortable with the concept of revisiting, reanalysing and reinterpreting data in response to evolving theory. In the case of this study, it was the qualitative interpretation and reconstruction of the quantitative data and results, within an overall grounded theory method, which allowed the data to answer the question – one of the most important components of rigorous research. It is surely a less rigorous approach to test every variable, in every possible combination, in the hope that you will chance upon a serendipitous result, rather than asking particular questions within the context of a known or emerging model.

It is important to realise when undertaking integrated analysis that the two strands of analysis, the qualitative and the quantitative, and their blending have particular rules. As in the construction of DNA, only certain sequences are possible, and only particular proteins (data) can bind together. In the case of this study, qualitative data were only bound to and analysed with quantitative data where they satisfied the 'rules' such as sample size, normal distribution, multiple response and so on. Quantitative data, when bound to qualitative data, served to inform and give insights but were not allowed to force a reductionist attitude to the qualitative analysis.

But finally, I would argue that such integrated methods are very rigorous because, ultimately, you cannot force any part into the organic helix – each component in the sequence has a key and it has to fit in place. In such a study, everything must 'fit' and 'work' – the rigour and the validity can then be judged by the functionality of the resultant organism. For the persons with spinal injuries who have reviewed the findings, the final model or community service need derived in this study is a very functional organism.

Writing up integrated analyses

I will now briefly turn to one of the less well travelled journeys through words and numbers, the journey of writing up an integrated analysis. Clearly the study could not be presented in the more traditional 'Introduction, Hypothesis, Method, Results, Discussion' framework. Typically, multi-method studies are written in two ways:

• The quantitative data is predominantly presented with the qualitative data (juicy quotes) being used to illustrate and humanise the points being made, or
• The qualitative data is predominantly presented with the quantitative data (statistics) being used to validate and bolster the credibility of the qualitative findings.

Neither of these methods of writing could accurately represent the organically integrated method used in this study. By writing in a manner consistent with the double helix metaphor of DNA it was possible to carry readers on the journey through words and numbers to understanding.

To do so, the processes and findings were presented in a series of iterative chapters. In each iteration, the 'sense' strand (whether derived from words or numbers, or indeed, an integration of both) was presented. The anti-sense strand would then be presented in counterpoint. Each iteration would then conclude with a discussion of the mutation believed to be required unwind the analysis and undertake reanalysis to better join the two strands into a cohesive organism:

sense à anti-sense à mutationà unwind à sense à anti-sense à mutation à unwind …

The anti-sense strand in each iteration (five in total) was 'built' around four case studies that wove the path through the thesis, providing a coherent framework for the reader. It is important to note, however, that whilst this form of presentation worked well in the presentation of the doctoral dissertation which is the subject of this paper, ways of translating this to the much more constrained writing style of journals have not yet been investigated.

Conclusion

It has not been my intention today to force upon you yet another methodological metaphor, DNA, but rather to argue that answers to complex questions can not be arrived by anything other than a mutating, qualitative method which can juxtapose numbers and words. I further argue that it is both possible and essential to maintain the rigour of the individual components within the overall method.

In this study, an overall qualitative approach allowed an understanding of what was needed as determined through the perceptions of persons with spinal injuries, as expressed through quantitative data and qualitative data. This thorough and rigorous approach ultimately produced a fresh insight and new approaches to the meeting the community service needs of the spinal injured population of New South Wales.

References

Begley, CM (1996) Using triangulation in nursing research. Journal of Advanced Nursing, 24(1), pp.122-128.

Billings, JR & Cowley, S (1995) Approaches to community needs assessment: A literature review. Journal of Advanced Nursing, 22(4), pp.721-730.

Bradshaw, J (1972) The concept of social need. New Society, March, pp.640-643.

Denzin, NK (1978) The Research Act: An Introduction to Sociological Methods. New York NY: McGraw-Hill.

Endacott, R (1997) Clarifying the concept of need: A comparison of two approaches to concept analysis. Journal of Advanced Nursing, 25(3), pp.471-476.

Kritzer, HM (1996) The data puzzle: The nature of interpretation in quantitative research. American Journal of Political Science, 40(1), commencing p.1.

Magidson, J (1993) SPSS for WindowsL CHAID, Release 6.0. Chicago: Statistical Innovations Inc.

Microsoft (1997) Microsoft Encarta97 Encyclopedia. Microsoft Corporation.

Miller, D (1976) Social Justice. Oxford: Clarendon.

Proctor, S (1998) Linking philosophy and method in the research process: the case for realism. Nurse Researcher, 5(4), pp.73-90.

Stainton, T (1994) Autonomy and Social Policy. Vermont: Ashgate Publishing Company.

Thayer, R (1973) Measuring need in the social sciences. Social and Economic Administration, 7(2), pp.91-105.